When one sample is inspected by optical method, the sampling points of the sample are normally consisted of many different and non-fixed points rather than one single point. Due to the absence of any regular feature points available on the sample for reference, the actual addresses of the sampling points may not be obtained by only using an open-loop motion method for sampling. Furthermore, if image capturing or signal detecting by using an optical scanning device equipped with a position feedback sensor, such as a laser galvo mirror equipped with a galvanometer or an optical encoder or a magnetic encoder, the position information of the optical scanning device is feedback. Then, the actual addresses of the sampling points on the sample are calculated by the feedback position information of the optical scanning device and the non-linear and complicated coordinate transformation formulas.
The said addresses of the sampling points are different and separated from the said feedback position information of the optical scanning device by a distance being far larger than the scale of the sampling displacement. Consequently, the position errors are amplified in the coordinate transformation, and result in differences between the calculated values and the actual addresses of the sampling points. Besides, for a live sample which needs to be periodically observed over a period of time, once the slide is removed from the equipment and then replaced to the equipment again, position change and image offset will occur.